PDF scan to USB stick

"

CROSS-NATIONAL COMPARATIVE STATUS...ATTAlNMENT RESEARCH

DonaId J. Treiman and Harry B. G. Ganzeboom

ABSTRACT Cross-national comparative studies of tbe process of occupational attainment conducted over tbe past 20 years are reviewed. Some support is found for tbe claim tbat industrial societies are more open tban developing societies, but substantive eonclusions regarding societaI similarities and differences in status-attainment processes are very limited and very tentative. This unhappy state is attributed to a failure by the research community to take issues of measurement comparability seriously. A proposal is made for a collective effort to produce standardized resuits from individuaI national studies, in order to facilitate mture cross-nationaI comparative research on status-attainment processes and pattems.

INTRODUCTION More than 20 years ago. Duncan (Duncan and HOOg.e 1963; Duncan 1966; Blau and Duncan 1967) transfonned tbe study of the intergenerational transmission of social status from a nearly exclusive concern witb tbc bivariate relationship Research in Sodai Stratffication ad Mobliity, Volume 9, pages 105-127. Copyright © 1990 by JAl Press mc. All riglits ot reproduction in any torm reserved. ISBN: 1-55933-205-8

105

106

DONALD J. TREIMAN and HARRY B. G. GANZEBOOM

between the occupations of fathers and sons into consideration of the diverse paths by which various aspects of parentaI status are converted into advantages and disadvantages for their offspring. By reconceptualizing status attainment as a process in which successive positions in the stratification system (education, first job, subsequent jobs, and income) are linked through a series of direct and indirect causal paths, and by demonstrating how the process could be modelled by means of path analysis (structural equations), Duncan established a paradigm that has been widely followed in the intervening years. The result has been a substantial elaboration of the initial model and cumulation of knowledge about status-attainment processes, at least within single countries-mainly the United States, but other countries as weIl. These developments have been reviewed in a number of artieles (HalIer and Portes 1973; Kerckhoff 1976; Mayer 1979; Matras 1980; Bielby 1981; Simkus 1981; Campbell 1983; Goldman and Tickamyer 1984; Kerckhoff 1984; for a critical review, see Colclough and Horan 1983 and the literature cited therein). From the outset, the main thrust of this research tradition has been comparative (although comparisons have been restricted for the most part to subpopulations within single societies). Very early on, it was recognized that the parameters of a status-attainment model could be read as describing a system of social stratification and hence that comparisons across social systems of magnitudes of such parameters could be infonnative about the nature of social stratification in different societies. Thus, modest (two or three country) cross-national comparisons of status-attainment processes began to appear early on (Balan 1968; Jones 1971; Iutaka and Bock 1972; Wilson 1972), as weIl as programmatic calls for more ambitious efforts to generate comparabIe status-attainment models for a large number of societies and then model the parameters themselves by reference to exogenous macrosocial variables (Treiman 1970, p. 214). The substantive research agenda of comparative status-attainment analysts was not new. Hypotheses were mainly derived rrom earlier theoretical writings (e.g., Lenski 1966, who blends together insights from traditional functionalism, Weberian conflict theory, and the study of technological evolution) or were borrowed from comparative research on intergenerational occupational mobility. Prominent among these is the thesis of industrialism (Treiman 1970), which posits that the level of direct status transfer from parents to their offspring will decrease with industrialization. A corollary is that as the level of average educational attainment increases (which generally accompanies industrialization), the influence of social origins on educational attainment wiII decrease. (However, others, e.g., Boudon 1973, have suggested the reverse possibility-that the increased length of schooling and the muItiplication of decision points in educational careers have actually increased the opportunities for high status parents to provide competitive advantages for their children.) Within the category of technology-related hypotheses is KeIley's (1978) claim that the more unequal the distribution of income (and wealth) and educational attainment in the population

I ~

~ t. ~~

~

à~

~

f

I I

I i

J

II ~

~

, f

t

I

Comparative Status Attainment

107

(of parents), the more influential parental status will be in detennining the educational and occupational careers of the offspring. Of more traditional functionalist origin than these propositions about technological factors and the unequal distribution of human resources is a hypothesis at the heart of the reasoning of Blau and Duncan (1967, pp. 429-431); they propose thata value change toward increased universalism as societies industrialize results in a shift from ascription to achievement as a basis for status attainment and a shift from direct occupational transmÎssion to indirect transmission through educational achievement. In a modem variation on this sort of value hypothesis, IngIehart (1977, 1981; cf. De Graaf 1988) has proposed that the growth of postmaterialist values in the Western world since the 1960s has decreased achievement motivation and has, therefore, reduced the strength of the connection between educational attainment and income attainment. All of the above hypotheses are variations on a theme-the effects of societal development. A somewhat different set of hypotheses is concerned with the influence of political structures, independent of the level of development. One important hypothesis is that intergenerational ties-both the dependence of educational attainment on social origins and the direct transmission of occupational status-should be weaker in state socialist or social democratie welfare states than in more laissez-faire capitalist states, because, in socialist societies, the state is more prone to use the educational system as a tooI for promoting equality of opportunity (parkin 1971). Heath, who also advances this argument, in addition proposes (1981, pp. 195-196,204-210) that mobility chances will be most limited under right-wing conservative regimes. In an interesting variation upon this political hypothesis, Kelley and Klein (1981) have argued that political revolutions reducè inherited advantage in the short run but that new elites, consisting of the educated sector of the ancien régime plus those who were able to exploit the revolutionary situation to become educated, soon emerge and move to secure their advantage for themselves and their offspring. Konrad and Szelenyi (1979) construct a similar argument with respect to Hungry after the socialist transfonnation. With so many interesting comparative hypotheses to be tested, and so much empirical work on specific societies, it would be reasonable to suppose that considerable progress has been made toward assessing the factors that affect status-attainment processes in different societies. Nearly 20 years after publication of the first comparative modeIs, however, our settled knowledge of whether, to what extent, and in what ways status-attainment regimes differ across societies is not much greater than it was in 1970. There are two main reasons for this. First, serious comparative research on social stratification, involving the systematic comparison of data from many countries, has engaged the energies of only a handful of scholars. For understandabIe reasons, most "comparative" studies have been restricted to binational comparisons of the scholar's own data and data from the United States or some

108

OONALO J. TREIMAN and HARRY B. G. GANZEBOOM

other "reference" society-or, at most, comparisons of three or four societies. But such comparisons have an inherent limitation-it is impossible to know which of the many ways that nations differ, if any, account for observed differences in their stratification systems. To the best of our knowiedge, only three studies (Kelley 1978; Sharda, Conaty, and Miller 1983; Treiman and Vip 1989) have attempted to systematically test the effects of societal contexts on status attainment by modelling coefficients ofmicro-models using macro-indicators. Second, there has been great resistance to the need for measurement comparability. But, as has been suggested previously (Treiman 1975, 1977; Krymkowski 1988), without comparability of measurement, it is very difficult to know to what extent societal differences in the values of nominally comparabie coefficients of statistical models reflect true differences in social structure and to what extent they merely reflect "instrument effects," that is, differences due to the way variables are measured-and, conversely, to what extent similarities in the values of nominally simiIar coefficients mask true differences in social structure. The consequence is that few generalizations regarding cross-national differences or similarities in the status-attainment process' can be regarded as definitivelyestablished. Nonetheless, a number of studies have compared the statusattainment process in two or more countries. In what follows, we pursue four tasks. First, we review these studies from the point of view of measurement comparability, to substantiate the strong claims we have just made regarding the lack of established knowledge and the reasons for it. Second, we attempt to extract from the corpus of comparative studies some generalizations that, if not unequivocally established, might at least be regardedas provisionally supported. Third, we propose a short-term solution to the problem of standardization-a set of standard analyses to be carried out by individual researchers in order coIlectively to create a sufficient body of standardized results to permit valid crossnational comparisons. FinaIly, we review new developments in status-attainment research in single countries that appear to us to be particularly promising for pursuit in a comparative framework. By status attainment, we mean the attainment of education, occupation, and income. If taken in a broad sense, this topic covers a huge body of literature, not only in sociology, but also in economics and education. To make our review of comparative studies manageable, we restrict it to intergenerational studies, that is, to studies in which status attainment is explained, at least in part, by parental characteristics; to studies that compare two or more nations; and to studies that involve at least three variables. However, we exclude the extensive literature, in both sociology and economics, on income attainment; existing evidence suggests .that direct parental influences on offspring's income are smallor nonexistent (Ouncan, Featherman, and Ouncan 1972, p. 40). Furthermore, we will not be concemed with thè extensive literature analyzing the bivariate relationship between father's and son's occupation (for recent reviews see Kerckhoff 1984; Kurz and Mueller 1987; Ganzeboom, Luijkx, and Treiman 1989). Finally, we will not

Comparative Status Attainment

109

be concemed with gender, ethnic, or racial differences in status attainment within single countries or with such matters as career mobility, except insofar as career mobility is affected by parental status (and is studied comparatively). While of considerable interest, all these topics would excessively broaden the scope of our review.

MEASUREMENT COMPARABILITY IN COMPARATIVE STUDIES OF STATUS ATTAINMENT To our knowIedge, there have been 28 cross-national comparisons of the process of occupational attainment, 10 of which have involved comparisons of more than two nations. 1 Table 1 summarizes these studies. The table shows standardized coefficients for a model predicting occupational status from father's education and occupational status, respondent's education, and the status of his (or her) first job-or for simpIer models omitting one or more variables-together with information about the sample and the coding of the variables. 2 Inspecting the tabie, it is evident that many analysts use the same data. But, even when different analysts use the same data, they do not always get the same results (compare, for example, the coefficients from the 1965 Australian study reported by Jones 1971; Featherman, Jones, and Hauser 1975; and Jones and McOonneIl 1977). Moreover, the situation is even worse if one attempts to generalize the findings from studies based on different data sets. There are several reasons for this. First, many analysts are quite cavalier about the properties of the samples being compared. Rural villages and towns in developing countries or small towns are compared with entire industrialized nations (e.g., Lin and Yauger 1975; KeIley, Robinson, and Klein 1981), cites are compared with nations (e.g., Holsinger 1975; Kelley 1978), married men (Iutaka and Bock 1972) or husbands of married women (Kelley 1978) are compared with all adult men, and men of different ages are c()mpared (e.g., Heath 1981). There has been littIe explicit evaluation of how strongly these noncomparabilities affect results, but, in some instances, the effects are likely to be large. Certainly, it is. well-known that the coefficients of status-attainment models differ sharply depending upon whether the farm-origin population is excluded or included (Featherman and Hauser 1978) and also for different age groups (Ouncan et al. 1972). To make thlngs worse, nominal identity does not ensure functional equivalence. lt might weIl be suggested, for example, that the specification of an identical age range (e.g., 20-64) in a comparison of occupational status attainment in Brazil, Japan, and the United States does not achieve true comparability in the populations being studied because a substantial fraction of Japanese and Americans in their early 20s are not yet in the labor force while virtually all Brazilians are and, on the other hand, since 55 is the standard retirement age in

Table 1.

Standardized Coefficients for the Regression of Status of Current Occupation on Status of First Job, Education, Status of Father's Occupation, and Father's Education, Cross-National Comparisons

Author

Country (fear)

Balan (1968, p. 185)

Mexico (1965) United States (1962)

Fe.thennan, Jones and H.user (1975, pp. 345-346)'

Australia (1965) Australia (1965) Australia (1965) United States (1962) United States (1962) United States (1962)

Heath (1981, pp. 218-219)

..... ..... o

Herz (1983, pp. 211-216)

Herz (1986, p. 122)

Holsinger (1975, p. 273)

Hope (1984, pp. 26-27)8

lutal